Bulletin
of
the
Seismological
Society
of
America,
Vol.
74,
No.
5,
pp.
1951-1968,
October
1984
ARRAY
ANALYSIS
OF
THE
GROUND
VELOCITIES
AND
ACCELERATIONS
FROM
THE
1971
SAN
FERNANDO,
CALIFORNIA,
EARTHQUAKE
BY
HsuI-LIN
LIU
AND
THOMAS
HEATON
ABSTRACT
Profiles
of
ground
velocity
and
acceleration,
displayed
as
a
function
of
epicen-
tral
distance,
are
analyzed
for
recordings
of
the
1971
San
Femando
earthquake.
Three
long
profiles
(>50
km)
and
three
short
profiles
(<2
km)
are
studied.
Although
there
is
considerable
variation
in
waveforms
and
peak
amplitudes
observed
along
the
long
profiles,
there
are
also
many
examples
of
coherent
phases
seen
on
adjacent
stations.
There
are
striking
differences
in
the
amplitudes
and
durations
of
ground
velocity
observed
at
stations
located
on
hard
rock
sites
as
opposed
to
stations
located
within
the
large
sedimentary
basins
of
the
Los
Angeles
area.
Furthermore,
the
San
Femando
Basin,
which
is
adjacent
to
the
source
area,
seems
to
respond
quite
differently
from
the
Los
Angeles
Basin
which
is
about
30
km
from
the
earthquake
source
area.
Ground
acceleration
profiles,
however,
show
that
there
is
no
corresponding
change
in
the
duration
or
amplitude
of
high-frequency
shaking
with
site
characteristics.
We
infer
that
the
excitation
of
surface
waves
within
sedimentary
basins
is
the
reason
that
large
peak
velocities
and
displacements
are
observed
for
soft
sites.
The
ground
velocity
waveforms
are
nearly
identical
along
the
three
short
profiles,
which
are
all
located
within
the
Los
Angeles
Basin.
Greater
variation
of
waveforms
and
amplitudes
are
seen
for
ground
acceleration
along
these
short
profiles,
although
strong
phase
coherence
is
still
observed.
INTRODUCTION
The
purpose
of
this
study
is
to
characterize
the
nature
of
ground
motions
recorded
during
the
1971
San
Fernando
earthquake.
We
systematically
display
the
data
in
a
fashion
such
that
we
may
gain
insight
into
wave
propagation
and
earthquake
source
phenomena.
Fundamental
questions
that
we
wish
to
address
are:
what
is
the
phase
coherence
as
a
function
of
frequency
and
station
spacing?
What
are
the
major
phases
seen
in
the
records?
And
what
are
the
effects
of
variations
in
the
local
seismic
velocity
structure
(e.g.,
mountains
and
basins)?
A
satisfactory
resolution
of
these
questions
would
obviously
require
an
extensive
and
very
sophisticated
mod-
eling
study.
Instead,
we
have
simply
organized
the
data
in
a
way
such
that
simple,
but
hopefully
meaningful,
interpretations
can
be
inferred
about
the
nature
of
the
ground
motions
produced
by
this
earthquake.
In
a
study
similar
to
this
one,
Hanks
(1975)
analyzed
the
ground
displacements
from
the
1971
San
Fernando
strong
motion
data
and
demonstrated
that
there
is
strong
coherence
from
station
to
station
in
the
longer
period
waveforms.
He
also
identified
direct
shear
phases
and
subsequent
surface
waves
at
certain
ranges.
In
this
study,
we
extend
Hanks'
study
(1975)
by
examining
shorter
period
waveforms
as
represented
by
the
ground
velocities
and
accelerations.
Ground
velocity
profiles
are
displayed
together
with
profiles
of
topography
and
geologic
structure
as
inferred
by
Yerkes
et
al.
(1965).
We
identify
possible
wave
types
and
discuss
the
effects
of
geologic
structure
on
the
amplitude
and
duration
of
the
recorded
ground
motions.
Given
the
distribution
of
accelerometers
that
recorded
the
San
Fernando
earth-
quake,
it
is
possible
to
construct
ground
motion
profiles
along
three
azimuths.
1951
~OTI
eFl04
Ep
3G
114
(•)
EPICENTER
OGII~
StTES
NOT
IN
STUDY
•
GIlO
ACCEL
EROGRAPH
MAJOR
FAULTS
EXPOSED
BASEMENT
ROCK
BASEMENT
ELEVATION
IN
THOUSANDS
OF
FEET
SEA
LEVEL
DATUM.
VlCTO~VI
LL£
SANTA
SUSANNA
MTS
00XNARD
SAN
SANTA
MON/CA
MTS
v~//
u/~~-rG:~,
~-~"
ii
SAN
GABRIEL
-.~l,,~um
z
~
"~
,,-;z.,..ll^
^^~
r
^^,..
V,4L
I~E
Y
OAREA
I
@AREA
2
@AREA
5
&az~#,
<x
"'~",
~
]-l'~M;2"~:
E075
C054
I)
059
~1
,
"',
"-4
'
I
("-
\'"
~'~
BASt~
e~3
r089
~249
~:lJ;,J,~
i
",-,
,,|0
D06s
r
o~8
i
J34
....
P217
G
112
l
131
~,
eFO87
$265
I~
15';'
N
188
"",,
P220
S766
K
159
J
i~e~~
~3
PACIFIC
~
OCEAN
SANTA
SAN
C;ATALIN
~
."
0
ON
TAR]O/.~
I
,,,
,,:,
o
,:
',
--'4.
0
E'
"~
oN
195
FIC.
1.
The
long
profiles,
I,
II,
and
III,
and
site
locations
of
accelerometer
recordings
of
the
1971
San
Fernando
earthquake
(modified
from
Hanks,
1975).
34
°
07'
I
34oo5'1-
54
°
03'
tN
HC
BEVERLY
HILLS
II31
I(34
"
S266/
/E083
~
j
D059
~
DO6
k
34ooi
'
DOWNTOWN
LOS
ANGELES,AREA
ACCELEROGRAPH
SITE
0
I
2
(KM)
I
I
,J
ANGELESI
1
II
8
°
15.OO'
I
118°
20.00
'
ARRAY
ANALYSIS
OF
THE
SAN
FERNANDO,
CA,
EARTHQUAKE
1953
Figure
1
shows
the
station
distribution
and
our
corresponding
profiles.
All
profiles
originate
in
the
epicentral
region.
Profile
I
extends
65
km
southward
across
the
San
Fernando
and
Los
Angeles
basins
to
a
station
on
the
Palos
Verdes
Peninsula.
Profile
II
extends
95
km
S40°E
along
the
front
of
the
San
Gabriel
Mountains
and
then
across
the
San
Gabriel
and
Los
Angeles
basins.
Profile
III
runs
90
km
N40°W
across
the
San
Gabriel
and
Tehachapi
mountains.
Profile
I
is
identical
to
Hanks'
TABLE
1
S!
MINUS
TRIGGER
TIMES
Station
Time
(sec)
Station
(#FI.)
Time
(sec)
Azimuth
I
Local
Array
1
C041
3.0
D065
(11)
0.0
C048
1.6
E075
(11)
1.3
D057
1.1
E083
(7)
0.8
D068
0.0
J148
(17)
5.6
Hl15
4.0
P217
(12)
0.8
Hl18
0.0
$265
(31)
5.6
J145
1.6
$266
(21)
5.6
L168
1.3
Local
Array-2
N191
0.0
C054
(39)
1.3
$262
1.9
F089
(8)
5.1
$267
6.1
F098
(8)
4.5
Azimuth
II
Gl12
(43)
1.9
Gll0
1.3
K157
(16)
2.7
G108
2.1
K159
(8)
4.3
G108
4.3
R253
(10)
5.6
H121
5.6
Local
Array
3
M180
0.0
D059
(19)
0.0
F087
0.0
I131
(10)
5.6
N186
0.0
I134
(15)
6.1
P220
0.0
N188
(16)
5.6
Azimuth
III
R249
(27)
5.1
E071
0.0
F102
0.0
F104
0.0
J142
5.0
J143
0.0
J144
0.0
M179
0.0
#
Building
height
in
terms
of
floors.
(1975)
profile
4,
and
profiles
II
and
III
are
somewhat
abbreviated
versions
of
Hanks'
profiles
1
and
2,
respectively.
The
San
Fernando
earthquake
was
also
well
recorded
by
strong
motion
instru-
ments
in
high-rise
buildings.
These
buildings
are
clustered
in
three
locations
on
the
north
side
of
the
Los
Angeles
Basin:
the
Miracle
Mile
area
of
Wilshire
Boulevard
(local
array
1);
downtown
Los
Angeles
(local
array
2);
and
Century
City
(local
array
1954
HSUI-LIN
LIU
AND
THOMAS
HEATON
0
(-
~=
i---i
1:3
o
o
03
~,
~--:
....
~
~,
.......
O3
~g.~_
~¢')
a.)
a_:~T
o-
E
o
0
0
0
"-"
l.o
0
~£
I/On
pi
o
,,04L,6,
~
__~
I
/
I,._
i
~01
il
.
i.,,8
------w~/k/~mv'J~k~
=r
i
I
I
i
E
0
~
O~
Fro.
3.
Transverse,
radial,
and
vertical
components
of
ground
velocities
along
profile
I.
The
corre-
sponding
free
surface
and
basement
surface
are
shown
to
the
left.
Dashed
lines
crossing
30
and
40
km
indicate
the
possible
phase
arrivals
of
surface
waves.
3).
Station
locations
for
these
dense
local
arrays
are
shown
in
Figure
2.
These
local
arrays
correspond
exactly
with
Hanks'
(1975)
local
arrays
1,
2,
and
3.
Time
histories
of
ground
velocity
and
acceleration
are
all
taken
directly
from
reports
published
by
the
Earthquake
Engineering
Research
Laboratory
at
Caltech
(1974).
Ground
motions
are
rotated
into
radial,
transverse,
and
vertical
components
and
are
then
displayed
as
functions
of
time
and
distance
from
the
epicenter
reported
by
Allen
et
al.
(1973).
Since
absolute
time
is
not
available
for
any
records,
some
assumptions
must
be
made
in
order
to
correlate
phases
from
one
station
to
another.
We
shifted
records
such
that
the
apparent
first
shear-wave
arrivals,
which
are
named
S!
by
Hanks
(1975),
are
aligned
vertically
in
the
profile.
S!
minus
trigger
times
are
listed
in
Table
1
for
all
stations
used
in
this
study.
Unfortunately,
in
some
cases,
the
identification
of
S!
is
very
difficult,
and
thus
some
of
the
more
distant
ARRAY
ANALYSIS
OF
THE
SAN
FERNANDO,
CA,
EARTHQUAKE
1955
b
R
V
0
20
40
60
0
20
40
60
l
i
t
l
,
i
i
i
0
5'
IO
1,5
ZO
25
30
35
4O
45"
50
55
60
65
(KM}
C041
.....
~
120.6
C041
---~
58.3
5i.5
JI45
HI
1,5
----~~
LI66
11.9
D068
.....
hill
li.6
15.
5
.....
I
--
16.5
HII8
.....
5.9
NI91
.....
I
I
I
]
¢
.
,.
#~'A^
18.1
DO68--J~L~-~-
56
I
D057
--~
$262
--~~..rw--,
6.
I
i
2.2
FI(~.
3.
Continued.
stations
are
probably
not
aligned
on
the
first
shear-wave
arrival.
Once
the
profiles
are
constructed,
we
find
that
the
ground
velocity
waveforms
are
surprisingly
coherent
from
station
to
station,
allowing
us
to
estimate
apparent
phase
velocities,
to
identify
possible
wave
types,
and
qualitatively
separate
the
source
from
the
path
effects.
The
features
of
each
profile
will
be
described
in
more
detail
as
follows.
LONG
PROFILES
Profile
I.
The
velocity
traces
together
with
the
corresponding
free
surface
and
basement
surface
topography
profiles
are
displayed
in
Figure
3.
Velocity
traces
begin
with
the
S!
arrivals.
From
a
study
of
both
strong
motion
displacement
waveforms
and
long-period
teleseismic
body
waveforms,
Heaton
(1982)
reported
a
preferred
source
model
consisting
of
two
approximately
equal-size
events.
H1
is
the
hypocenter
of
the
first
event
which
ruptured
along
the
Sierra
Madre
fault,
and
4
1956
HSUI-LIN
LIU
AND
THOMAS
HEATON
sec
later,
rupture
initiated
on
the
San
Fernando
fault
at
the
second
hypocenter,
H2.
According
to
Heaton's
{1982)
interpretation,
the
high-peak
velocity
observed
at
station
C041
(Pacoima
Dam)
is
caused
by
rupture
directivity
along
the
Sierra
Madre
fault.
Although
the
faulting
process
is
rather
complicated,
the
total
source
duration
is
only
about
7
sec.
We
believe
that
the
signal
duration
at
Pacoima
Dam
(C041)
represents
the
approximate
duration
of
the
source.
0
~
5~
IO
I151.5
c
n'Ysec
2
I
\',
15
~
(
,
",,
0
--\
tfl~-
I
\
~
.202.0
20
C048
m.~.L~,.,,~.~.'~'~%"
'Y
t"
~
~'
'V
105.5"
.....
..,~
HII5
202.8
30~LI66
'T"~t'I'~"C'~-
~)~'"
-"
~
565
[
~o]
........
%e~"?/'~?¢
;~'''¢''~
~
%;!5
..........
I
45j,
5
"~/~
...........
,v~.,~pr~,~,~o~..~s:
''~
.........
--
55
Profile
[
radial
acceleration
60
NI91
56.7
(KM)
0
(0
2'0
50
(SEC)
FIo.
4.
Radial
components
of
ground
acceleration
along
profile
].
One
of
the
most
impressive
features
seen
in
Figure
3
is
the
correlation
of
ground
motion
waveforms
with
the
topography
of
the
subsurface
basins.
There
are
three
stations
(C048,
J145,
and
Hl15)
located
within
the
San
Fernando
Valley.
The
waveforms
recorded
at
these
stations
are
poorly
correlated
and
the
signal
durations
ARRAY
ANALYSIS
OF
THE
SAN
FERNANDO,
CA,
EARTHQUAKE
1957
are
about
30
sec.
These
durations
are
much
longer
than
those
seen
at
neighboring
and
more
distant
stations
located
near
the
Santa
Monica
Mountains,
such
as
L166,
D068
and
D057.
This
contrast
can
be
seen
in
Figure
3.
Coherent
waveforms
begin
to
appear
at
station
L166
and
signal
durations
drop
to
about
10
sec.
Furthermore,
peak
amplitudes
drop
by
about
one-half
after
passing
the
San
Fernando
Valley
stations.
Amplitudes
increase
again
at
stations
D057
and
$262
when
the
waves
reach
the
Los
Angeles
Basin.
The
waveforms
are
quite
coherent
in
the
beginning
6
sec
and
there
appears
to
be
little
moveout
from
the
S!
phases.
The
dashed
lines
passing
stations
L166,
D068,
D057,
and
$262
indicate
the
possible
arrival
of
surface
waves,
which
are
characterized
by
gradual
moveout
from
the
S
phases.
They
appear
dispersive
in
nature,
and
Hanks
{1975)
demonstrates
that
the
radial
and
vertical
components
of
ground
displacement
are
dominated
by
retrograde
elliptical
particle
trajectories
at
these
stations.
At
ranges
of
50
km
and
beyond,
it
is
hard
to
identify
the
body
waves,
and
the
waveforms
appear
to
be
well
dispersed
and
are
probably
composed
mainly
of
surface
waves.
The
apparent
moveout
velocity
of
the
surface
wave
phase
from
the
S!
phase
is
estimated
to
be
5.7
km/sec.
If
we
assume
that
the
phase
velocity
corresponding
to
the
S!
phase
is
3.5
km/sec,
then
the
phase
velocity
of
these
surface
waves
is
about
2.2
km/sec.
It
is
useful
to
contrast
the
nature
of
wave
propagation
for
waves
observed
in
the
San
Fernando
Valley
and
the
Los
Angeles
Basin.
Peak
ground
velocities
observed
in
both
basins
are
high
with
respect
to
sites
in
the
Santa
Monica
Mountains.
Although
long-period
surface
waves
are
clearly
visible
in
both
basins,
the
duration
of
these
surface
waves
is
up
to
20
sec
longer
in
the
San
Fernando
Valley,
the
basin
which
is
closer
to
the
earthquake
source.
The
earthquake
ruptured
into
the
northern
part
of
the
San
Fernando
Valley
which
is
a
closed
sedimentary
basin.
We
hypoth-
esize
that
the
long
signal
durations
are
caused
by
surface
waves
which
are
laterally
trapped
within
the
valley.
The
short
signal
durations
seen
at
adjacent
stations
in
the
Santa
Monica
Mountains
indicates
that
these
reverberating
waves,
which
developed
in
the
San
Fernando
Valley,
did
not
propagate
away
from
the
valley.
The
relatively
simple
waveforms
observed
in
the
Santa
Monica
Mountains
are
probably
composed
primarily
of
shear
body
wave
arrivals.
Whereas
peak
velocity
amplitudes
in
the
San
Fernando
Valley
are
associated
with
surface
waves,
the
absence
of
these
surface
waves
in
the
adjacent
Santa
Monica
Mountains
is
probably
responsible
for
the
noticeable
drop
in
the
peak
amplitude
of
velocity
that
we
observe
as
we
proceed
from
the
San
Fernando
Valley
into
the
Santa
Monica
Mountains.
As
we
proceed
further
southward
into
the
Los
Angeles
Basin,
a
new
surface
wave
train
appears
to
develop
and
peak
velocities
jump
higher
again
with
respect
to
those
seen
in
the
Santa
Monica
Mountains.
It
appears
that
the
Santa
Monica
Mountains
constitute
a
significant
barrier
across
which
the
surface
waves
that
developed
in
the
San
Fernando
Valley
and
Los
Angeles
Basin
did
not
propagate.
In
Figure
4,
we
show
the
radial
components
of
the
ground
acceleration
for
stations
located
along
profile
I.
Traces
begin
at
their
trigger
time
and
are
aligned
vertically
with
respect
to
the
S!
phase.
Although
there
is
considerable
variation
in
the
observed
waveforms
from
station
to
station,
the
duration
of
the
high-frequency
motions
is
a
fairly
constant
10
sec.
Furthermore,
strong
phases
can
be
seen
on
many
records
at
about
4
and
8
sec
after
the
initial
S
arrivals.
We
speculate
that
the
overall
duration
and
timing
of
arrivals
seen
on
the
relatively
high-frequency
acceleration
time
history
are
directly
related
to
the
details
of
the
faulting
process.
It
is
suggested
that
the
basin
geometry
has
less
effect
on
these
higher
frequency
waves.
1958
HSUI-LIN
LIU
AND
THOMAS
HEATON
T
o
O(sec)
20
40
60
'
'
'
'
T
/
o
¢n
zo
.........
~Io,o~--~
~
~
'°'~
L
40JHI21
--~
19.9
[
._c
o
,,,
~n
1
I~
~
7
00
L
0
~
I
Io
rr
~
\..J~
/i7,,o.
,o
I
,~
,,~__.~,~a~.~
~.~
/l t
°
I
I
I
I
l
J
J
~
~
J
i
i
l
II
O
h
0
0
E
0
i~
0
OE
0
_~
--
0
"-"
t,O
-~
FIG.
5.
Transverse,
radial,
and
vertical
components
of
ground
velocities
along
profile
II.
The
corre-
sponding
topography
and
the
basement
surface
are
shown
to
the
left.
Dashed
lines
indicate
the
possible
phase
arrivals
of
surface
waves.
Profile
II.
Profile
II
runs
along
the
southwest
border
of
the
San
Gabriel
Mountains
and
then
extends
across
the
San
Gabriel
basin
and
into
the
Los
Angeles
Basin.
The
velocity
traces
(beginning
at
S!)
together
with
the
path
profile,
are
shown
in
Figure
5.
Somewhat
different
features
are
observed
along
this
profile.
Velocity
waveforms
are
relatively
simple
and
have
durations
comparable
to
the
source
duration
for
stations
within
40
km
of
the
source
region.
Sediments
are
either
thin
or
absent
for
these
stations,
and
much
of
the
waveform
appears
to
be
comprised
of
shear
body
ARRAY
ANALYSIS
OF
THE
SAN
FERNANDO,
CA,
EARTHQUAKE
b
0
5"
I0
15'
20
25.
5(3"
55'
40-
45-
50-
55-
60"
65
~
70"
7'5-
80
e5
9o
(KM)
R
V
0
20
40
60
0
20
40
60
I
I
I
i
C04t
---.~
120.6
I
I
14.7
GItO
--"~1
G,O6
__~~
GI08--~;
-
-
t3.7
'1
HI21
--,1~,~
II
.2
I
I
I
I
NI86
--~
I
I
co,,
I
I
G,,O
--,~lvv,.--~
$9
I
GL06
GI08
8.7
HI21
--
~-v,v~
8,2
I
I
I
I
I
HI24
--~
4.3
5.9
H124
--
r"V~AJ
2,3
MI80
--~
7.5
F087
--
P220
--~
I
I
I
I
5.9
MI80
--~
2.4
:087
--~
5.5
a220
--~
k
i
I
I
FIG.
5.
Continued.
1959
waves.
However,
Heaton
and
Helmberger
(1979)
show
that
a
nondispersed
funda-
mental
Rayleigh
wave
may
comprise
a
significant
part
of
the
longer
period
motions
at
these
stations.
The
dashed
lines
covering
the
range
from
30
to
50
km
shows
appreciable
moveout
of
a
later
long-period
phase
with
respect
to
the
initial
S
arrivals.
If
we
assume
a
horizontal
phase
velocity
of
3.5
km/sec
for
the
S!
phase,
then
the
phase
velocity
corresponding
to
the
dashed
line
is
approximately
2.7
km/
sec.
This
is
higher
than
the
phase
velocity
inferred
for
surface
waves
identified
in
the
Los
Angeles
Basin
for
profile
I.
This
observation
is
consistent
with
the
fact
that
sediment
layers
are
relatively
thin
compared
with
those
encountered
along
pro-
file
I.
Ground
velocities
at
stations
that
lie
in
the
Los
Angeles
Basin
at
distances
of